This slim volume (which might perhaps more understandably bear the main title From Syntax to Phonology) packs a lot of claims, evidence, and argumentation into its 112 text pages. Starting with 20 well-known cases from Otfrid's Evangelienbuch where primary umlaut appears to operate across word boundaries (as in nem iz for an expected nam iz ‘took it’), the author argues in chapter 2 (Prosodic deficiency and cliticization in Otfrid) that they indeed represent instances of cliticization, and that this cliticization is a synchronic process in the language of Otfrid (meaning that these forms are not lexicalized as Janda (1998) would have it, but rather productively produced). This also involves recognizing primary umlaut as a productive, and indeed postlexical, phonetic-phonological process for this language.

In order to do this, the author has to go into some detail to describe how she sees the productive process of cliticization as working. For her, the crucial environment for syllabification, cliticization, and umlaut, among a number of other processes, is the prosodic/phonological word, or pword, whose boundaries frequently, but not always, coincide with morphosyntactic ones. The author accordingly devotes a number of pages to the exact nature of the pword and its place in the prosodic hierarchy, following this up with the important question of what constitutes a minimal pword in Old High German. Based on work done in modern German, especially Hall 1999, but also in other old West Germanic languages (she mentions Dresher & Lahiri 1991 and Lahiri & Dresher 1999, among others), she establishes certain important principles governing the pword in Otfrid: A pword must consist of at least one foot, which necessarily comprises one stressed syllable; it is minimally bimoraic, but there is evidence that word-final CVC sequences are monomoraic; lexical words are pwords, but function words such as pronouns are not. Interestingly, these principles are sometimes in conflict; importantly for the author, some lexical words may appear as monosyllabic CVC, making them monomoraic, which in turn, the author argues, makes them susceptible to certain postlexical prosodic processes.

These postlexical prosodic processes lie of course at the heart of this book. Cliticization occurs, as the author sees it, when a prosodically deficient pronoun is postlexically assimilated to the pword formed by a (normally moraically challenged) lexical (or in the case of modals also function-word) host. In chapter 2, the hosts are all verbs, whether lexical or functional, and the focus is on the apparently exceptional examples of primary umlaut to be found on the host. Chapters 3 and 4 broaden the argument to other forms found in Otfrid, for which the author argues that a similar analysis is appropriate.

In chapter 3, the forms involved are theih, theiz, and theist, clearly related in some way to the also-attested forms thaz ih ‘that I’, thaz iz ‘that it’, and thaz ist ‘that is’. These (similarly umlauting) forms are widely documented throughout the Evangelienbuch, and indeed in most cases outnumber their non-cliticized kin. Also interesting is that these forms, or actually their descendents deich, deiz, and deist (dêst, dest) are documented in Middle High German, unlike the forms of chapter 2. In her analysis of these data, the author must to a certain extent wade through a number of terminological issues. Thus on p. 33 she notes Janda's characterization of forms like these as “lexicalized,” which she essentially finds to be insufficiently precise. Instead, she claims that they “should likely be discussed within the framework of grammaticalization.” But as a precondition for either lexicalization or grammaticalization, she finds it useful to speak of “form fossilization,” which she defines as the process whereby a [host + clitic] sequence becomes lexically listed (p. 34). The question she then emphasizes in chapter 3 is whether these at least originally productively cliticized forms had by Otfrid's time already become fixed units in the lexicon, or were still phonologically derived.

In chapter 2, the author had used the argument of relatively low frequency to find that forms like werf iz for expected warf iz ‘threw it’ were unlikely candidates as such “fossilized” forms. As she shows in a chart on p. 35, she cannot use the same argument for theih, theiz, and theist, which as a group outnumber their non-cliticized counterparts (approximately 60% to 40%, though most prominently in the case of theiz, least for theist). Yet she still finds these forms to be the products of active cliticization in Otfrid, based on two other important factors arguing against fossilization: (i) the cliticized forms always appear in the same syntactic positions as their non-cliticized counterparts; (ii) one or another, or both, of the two components contributing to the cliticized forms may have multiple syntactic roles (e.g. complementizer, relative pronoun, subject demonstrative, object demonstrative, etc.). Of course, in order to demonstrate the latter the author must give some idea of the syntactic framework she is operating with. Notably, a large part of chapter 3 (pp. 36–47) is devoted to outlining the author's views on Early Germanic and OHG clause structure (the title of section 3.2). Certainly the position taken there that the Otfrid text was written in the middle of a transformation from an older Germanic TP structural type to a later CP type is interesting in and of itself, but in this book it mainly serves the function of establishing the syntactic slots that the parts of clitics fill.

The author's arguments against the fossilization of these forms is most successfully made for theih, which is also the least documented, vs. its non-cliticized counterpart. It is also somewhat supported for theiz (the most documented vs. its non-cliticized counterpart), although here “syntactic role” depends rather more heavily on the author's syntactic framework. Theist, finally, even according to her syntactic framework, shows in all respects the most evidence of fossilization. She simply notes that “we will err on the side of caution and assume that we cannot count the theist forms among our still large group of actively-produced clitic groups” (p. 68).

To be really cautious, I think she should probably rule out theiz as well. She must concede that the thaz part of the form is always a complementizer, and the putative syntactic heterogeneity of the iz part relies in part on three examples she gives of object iz, but primarily on the distinction she makes between ergative subjects and unaccusative subjects within her syntactic framework. Since elsewhere (p. 55, note 12) she seems to find it important that an argument does not crucially hinge on her syntactic proposals, why not here as well?

Indeed, the author herself gives even more reasons to see theiz as fossilized in Otfrid. It turns out that in fact, while it is not categorical, the collocation thaz iz or theiz has an ergative subject as the second component about 85% of the time, and 90% of these instances show the cliticized variant! Not so far from the situation found with the successor form deiz in the Middle High German Parzival text, to which the author appears to grant fossilization status (p. 64). To characterize the theiz structures in Otfrid as “close to fossilization” (also p. 64) sweeps a real question under the rug: How does one handle a continuum in a grammar that wants a yes/no answer (is theiz listed separately in the lexicon or not?). This is not just her problem, of course.

In chapter 4 (Elision and cliticization) the author adds to the number of cliticized forms in Otfrid a further set of phenomena that are commonly ascribed to the requirements of meter in this poetic text. Thus the later focus in this chapter is on elisions such as wolt er for expected wolta er ‘he wanted’, but first she returns to theih, theiz, and theist, now to answer the question as to whether their usage in place of the non-cliticized forms might be metrically rather than prosodically motivated. In the course of her argument, the author outlines and illustrates the metrical types ascribed to Otfrid by Bostock (1976), and then uses them to explore this question. Her basic finding, enunciated already on p. 77, is that “the poetic meter, to a limited extent, does influence whether the cliticized or non-cliticized variant is attested.” After the consideration of a large number of verses, the author concludes (p. 87) that roughly half of the cliticization/non-cliticization choices might be seen as being influenced by the poetic meter. The rest, however, must be looked at prosodically.

The author makes a similar argument, with voluminous documentation, for the many examples of weak or preterite-present verbs which show, or do not show, elision of the final vowel in the preterite ending ‑ta before a function word beginning in a vowel. Roughly half of the choices made might be seen as due to the requirements of the meter, but the rest must be seen as prosodically motivated. And for the author, this clearly means cliticization as defined above. The chapter is rounded out, with similar argumentation, for other hiatus forms in which not the first, but the second vowel is deleted (zaltaz vs. zalt iz for zalta iz ‘said it’, zell ih vs. zelluh for zellu ih ‘I say’), and the author introduces further forms, such as brahta sa for expected brahta sia ‘brought her’, which may show us further manifestations of cliticization.

In her concluding chapter 5, the author not only summarizes the arguments of the preceding chapters as I have outlined them above, she also devotes a number of pages to a plea for scholars to return for their data to the original manuscripts in which the Otfrid text was written (as opposed to the later normalized editions). She also points to a number of phenomena in Middle High German and Middle Dutch that could be usefully approached with the framework she employs in this book. The book concludes with three appendices giving complete citations for the forms she mentions in the body of the text, and a list of references.

Certainly the arguments presented in this book are interesting and cogent, even if sometimes open to scholarly dispute. But this volume would also have profited from a careful proofreading. Obviously it went through a number of revisions after its appearance as a Wisconsin-Madison dissertation, but unfortunately some of the earlier versions seem to have left their traces in the published product. Thus in her chart on p. 32 the author appears to have found a few more examples of cliticized theih, theiz, and theist in the Evangelienbuch (namely 193) than she asserts to have found in the prose (also p. 32) preceding the chart (namely 189). I assume this reflects a recount done in a later perusal of the manuscript; unfortunately, it's not immediately apparent which is the more accurate number.

Similar text-historical causes probably lie behind the somewhat confusing table references on p. 8, where the author twice mentions table 4 found on p. 6 (with three attestations of the meg iz type), but is really discussing table 5, also on p. 8, with all twenty Evangelienbuch attestations, of which those found in table 4 form a subset. The same could be said of her reference, on p. 69, to “(24) on page 87 of this chapter”: there is no p. 87 in the relevant chapter, and table 24 of the chapter in question gives a different percentage for the phenomenon she is talking about than the text on p. 69. Other infelicities can also be found in the text, including missing letters and words, wrong word choices, and so on. None of this is unforgivable, but it's certainly distracting.